In toxicologic pathology, “blind slide reading” involves procedures in which the pathologist is unaware of the treatment group status of individual animals during the histologic slide evaluation phase of a toxicologic bioassay. Blind slide reading continues to be a controversial topic, despite published recommendations on this subject authored by professional societies and governmental regulatory agencies (e.g., The Society of Toxicologic Pathology and the US FDA). The reason for this is understandable.
The Why and How of Blinded Evaluations
From the earliest days of their careers, scientists and statisticians are cautioned repeatedly about the perils of sampling and observational bias, and correctly so, because it is undeniable that such factors have the potential to adversely impact study findings. Consequently, randomized sampling and blinding procedures have become axiomatic in contemporary scientific research, and experimental designs routinely incorporate such measures to varying degrees. Given this practice, the following question is often posed: “In terms of blinding, why should histopathology be handled differently from other endpoints?” This is a valid point of concern, but one that is easily addressed. Unlike most other types of assays, which typically generate a set of quantitative results that are not subject to optical recognition or experience-based interpretation, toxicologic histopathology involves the visual observation of morphologic differences in histologic sections from substance-treated specimens as compared to those of untreated controls. This observational step requires the pathologist to first discern such differences, some of which may be very subtle, among myriad complex microanatomic features that exist within each of the examined tissue types. Therefore, the initial goal of the histopathological assessment is to ensure, to the greatest extent feasible, that toxicologically-relevant findings are not missed, i.e., to guard against false negative (= Type II) errors. In order to best accomplish this objective, it is imperative that the pathologist be allowed to knowingly compare tissues from test subjects that are most likely to exhibit changes (i.e., those of the high dose group) to the tissues of concurrent negative controls. The importance of this step cannot be overemphasized, as there may be little future opportunity to detect findings that were missed originally (unless the slides are ultimately peer-reviewed). Once the initial examination has revealed all relevant findings, it becomes a simple matter for the pathologist to then mask and randomize slides that contain the target tissues, and re-assess the now blinded slides in order to confirm (or disprove, as the case may be) the existence of one or more previously recognized effects. This second phase of the histopathologic evaluation is equally critical to the first, as it prevents the pathologist from reporting findings that represent false positive (= Type I) errors. It is through this two-phase process, in which blinding only occurs after all important findings have been identified, that experienced toxicologic pathologists can successfully mitigate issues of observational bias, and thereby produce histopathology data that are both accurate and reliable.
The Disadvantages of Initial Blinded Evaluations
As described above, the primary disadvantage of blinded evaluation during the initial assessment phase is that one or more subtle treatment-associated effects may not be appreciated by the pathologist, which can lead to false negative study outcomes. But there are additional disadvantages. Although experienced anatomic pathologists typically have mental pictures of the “normal” appearance of the various organ types as viewed in histologic sections, they also recognize that each toxicological study is unique, in that the tissues from negative control animals in one experiment frequently differ in many aspects from those of control animals in the next study. Consequently, in blinded studies it can be challenging for the pathologist to distinguish potential treatment-related alterations from normal biological variation that exists among any group of animals. This is especially true for incremental types of changes, such as differences in cell or organ size, number, or coloration. Furthermore, if liver cell size, for example, is noticeably larger in one animal versus another, it can be difficult for the blinded pathologist to know which represents the normal baseline, and therefore decide whether one liver should be diagnosed as having hepatocyte atrophy, or the other as exhibiting hepatocyte hypertrophy. The end result of all this uncertainty is that blinded pathologists tend to conservatively diagnose many types of findings they would typically not record in non-blinded studies, because in the latter case they would be better able to appreciate the full range of normality. Consequences of recording numerous unnecessary diagnoses include increased slide evaluation time (which translates to increased expense and delayed reporting), and cluttering of the data, which in turn has the potential to negatively impact data interpretation.
When Are Blinded Evaluations Appropriate?
In addition to confirming the validity of treatment-related findings as described above, blinded histopathologic evaluations are appropriate for a variety of circumstances in which the characteristic toxicologic effects of a tested substance are already well-established through comprehensive prior experimentation, i.e., the odds of discovering further novel effects are either slim or non-existent. For example, blinded assessments can be extremely useful for reviewing lesion severity grading, to ensure that such scoring has been performed consistently throughout the course of a study. Initial blinded evaluations are also often suitable for efficacy studies, because those experiments tend to focus on the amelioration of specific known effects (however, for efficacy studies it is still advisable that the pathologist have unblinded awareness of the positive control group, if available). Blinding (in addition to randomized sampling and randomized order of evaluation) is also highly recommended for most types of morphometric analyses, to minimize the potential for subjective bias to influence individual measurements. Notably, Pathology Working Group (PWG) reviews are also usually conducted in a blinded manner, in part because the issues to be resolved are already clearly recognized. On the other hand, study peer reviews are typically performed non-blinded, because one important goal of this quality control procedure is to confirm that the original pathologist did not miss potential treatment-associated effects.
Blinded Evaluations and Medical Device Studies
Blind slide reading is prevalent in the medical device evaluation industry. Much of this is undoubtedly driven by the quantitative basis of engineering and manufacturing of devices to exact specifications. Medical device design teams often pull heavily from engineering and materials scientists who are used to precise measurements and quantitative data sets. However, blinded assessments of medical device studies may not always be advantageous. For the reasons enumerated above, the histopathologic analysis of medical device studies relies on the detection of observational differences between the interaction of the test device with the biological test system in comparison to a predicate device or sometimes a sham; to a large extent, blinded slide reviews negate such comparisons. Also, for some medical device studies, blinded evaluations may extend to the masking of various sacrifice intervals, so that the chronologic order of events is unknown to the pathologist. Pertinent changes may be lost in the background of findings that are reported, often the default when the toxicological pathologist reports all changes whether important to the test device and its timeline or not. In other words, the noise of many diagnoses may obscure subtle differences. Furthermore, many devices are unique in material, material composition percentage and structure. It is very helpful to the toxicologic pathologist to have a test sample of the device, either whole before processing to slide or on the slide, in order to have a full three dimensional idea of its potential interactions with the biological system under test. However, it is often the case that blind slide evaluation is requested without the provision of such test samples.